In systems in which a plurality of information processing apparatuses perform distributed processing, various techniques for performing appropriate load distribution have been proposed. In such distributed processing, not only allocation of information processing but also reallocation thereof may be performed on the basis of the operational statuses.
Examples of employed techniques for load distribution include a technique in which an agent selects an information processing unit with low utilization among a plurality of information processing units in a system performing distributed processing and allocates a processing request to the selected information processing unit. In this technique, each information processing unit transmits, on the basis of the processing status thereof, a request for reallocation of a processing request to the agent. Upon receipt of the reallocation request, the agent migrates a processing request managed by another information processing unit to the information processing unit that has transmitted the reallocation request.
According to another exemplary technique, an object-oriented client server system manages respective load statuses of servers and migrates a server object to a server with a low load on the basis of the load statuses.
Further, a technique has been proposed in which distribution of applications is performed in accordance with a predetermined algorithm when applications are executed in distributed platforms.
Japanese Laid-open Patent Publication No. 09-231184, Japanese Laid-open Patent Publication No. 2000-10936, and Japanese Laid-open Patent Publication No. 2004-520641 disclose related techniques.